Bayesian Based Random Weight Neural Network for Big Data Classification
STATISTICS, COMPUTING AND INTERDISCIPLINARY RESEARCH · 31.12.2025 · 7 (2) · 531-545
DOI: 10.52700/scir.v7i2.237 ↗Kurz gefasst
Fachartikel in der Zeitschrift STATISTICS, COMPUTING AND INTERDISCIPLINARY RESEARCH, online erschienen am 31.12.2025. Verfasst von Ishrat Khan und 2 weiteren Autor:innen.
Abstract (Originalsprache)
Random weight neural networks have shown significant performance in solving data classification and regression problems, primarily within the context of big data. However, randomization of input-layer weight and hidden-layer biases may affect the performance and generalization of extreme learning machines (ELMs). We propose the Bayesian-Based Random Weight Neural Network (BB-RWNN) a novel approach which improves the stability and generalization of RWNN by employing Attribute Bagging (AB) to train base ELMs on randomly selected attributes. BAB-RWNN incorporates sampling with replacement to randomly construct several attribute subsets. Each subgroup trains a base ELM. We achieve a final outcome by merging the ELMs via weighted averaging or voting. BAB-RWNN makes use of Naïve Bayes likelihood for weighting samples. The experimental results reveal that BAB-RWNN presents better prediction stability and accuracy when compared with conventional ELM and AB-ELM.
Zitieren
APA
Khan, I., Ashfaq, R. A. R., & Ali, A. (2025). Bayesian Based Random Weight Neural Network for Big Data Classification. STATISTICS, COMPUTING AND INTERDISCIPLINARY RESEARCH, 7(2), 531-545. https://doi.org/10.52700/scir.v7i2.237
Vancouver
Khan I, Ashfaq RAR, Ali A. Bayesian Based Random Weight Neural Network for Big Data Classification. STATISTICS, COMPUTING AND INTERDISCIPLINARY RESEARCH. 2025;7(2):531-545. doi:10.52700/scir.v7i2.237
BibTeX
@article{Khan2025,
author = {Khan, Ishrat and Ashfaq, Rana Amir Raza and Ali, Asghar},
title = {Bayesian Based Random Weight Neural Network for Big Data Classification},
journal = {STATISTICS, COMPUTING AND INTERDISCIPLINARY RESEARCH},
year = {2025},
volume = {7},
number = {2},
pages = {531--545},
doi = {10.52700/scir.v7i2.237}
}
EUVIMED